Typing single nucleotide polymorphisms (SNPs) is a key method of identifying risk variants associated with disease. Historically, the scope of genotyping experiments has been limited by the technology available. Gel electrophoresis-based genotyping methods are limited in sample- and SNP-throughput[1]. Developing these assays can often be labor-intensive, relying on the makeup and structure of the region surrounding the variant for optimization[1]. TaqMan genotyping assays, developed by Life Technologies, can run a large number of samples quickly and with minimal technician involvement[2], but SNP-multiplexing restrictions continue to limit the total number of genotypes to well under one million per day[3,4]. Sequenom's iPlex platform can also run many samples at once, but, as fewer than one hundred SNPs can be multiplexed together, throughput is comparably low overall[5]. Beckman Coulter's SNP stream technology could theoretically produce over three million genotypes per day, but this technology limits project range to a maximum of only forty-eight SNPs per reaction[4,6]. While the GoldenGate assay can process nearly two hundred DNA samples each day on hundreds or thousands of SNPs per sample, the price per genotype is not competitive with advanced, ultra-high-throughput techniques when typing over three thousand SNPs at once[4,7]. In order to process several million genotypes per day, the scale required for large genome-wide association studies, array-hybridization assays have become the most cost-effective option on the market.
Affymetrix's line of hybridization arrays and Illumina's line of Infinium-based arrays allow potentially hundreds of samples to be typed on hundreds of thousands or millions of SNPs in parallel[4,8]. These SNPs can be scattered across the entire genome, localized in regions of interests, such as exomes, or customized to the user's preference. These arrays have the benefit of not only being able to accurately genotype one million SNPs per sample at once, but also to measure copy number variation, potentially unveiling chromosomal abnormalities. Infinium-aligned OMNI BeadChip arrays currently have the ability to genotype up to nearly five million markers per sample, including half a million custom loci, on up to nearly one hundred samples each day.
As most diseases have a genetic component, these large-scale experiments can be crucial in finding genes associated with disease. High-throughput genotyping allows for efficient genotype generation in sample sets large enough to convincingly detect genetic association at lower minor allele frequencies. Whole-genome genotyping projects can be used to locate regions with statistically significant case-control allele frequency or copy number differences[9,10,11]. According to the National Human Genome Research Institute, genome-wide association studies led to 1,490 separate publications between November 25, 2008 and January 25, 2013, stemming from the discovery of 8,283 SNPs with a p-value less than 1 x 10-5 (see http://www.genome.gov/gwastudies/). These studies, which researched conditions ranging from height to testicular cancer, benefited from the broad approach afforded by a genome-wide analysis. In cases such as these, entire regions of interest might have escaped notice had the scope of typing been too restrictive. Thus, for large-scale association analyses, a genome-wide genotyping technique is the technique of choice.
Different versions of the Infinium assay exist, each intended for use with specific types of arrays. The InfiniumUltra assay, discussed in depth below, is appropriate for many 12- or 24-sample array chips. These often genotype over a hundred thousand SNPs per DNA sample and focus on targeted regions, such as on exome or custom panels. Other assay versions might be required for other chip types, such as the whole-genome genotyping arrays. However, as all Infinium assays share a common basis and mainly differ only by the reagent names, the reagent volumes, or the exact staining reagent procedure, techniques perfected on one assay version can often be universally applied. Other arrays, such as methylation arrays, might use a nearly-identical protocol, as well. Care must be taken to only use the version of the assay required for the chip type in use. Some types, such as ones measuring gene expression level, might require use of a nonInfinium protocol.
Samples must be processed in batches. For example, with the InfiniumUltra assay, prehybridization reagent tubes contain enough volume to run 96 samples, and the tubes cannot be refrozen. Therefore, samples must be run in batches of 96 samples at a time. The samples will be amplified on the first day. After approximately 1 hr of benchwork, the samples must be heated in a convection oven for 20-24 hr. The following day, nearly 4 hr will be spent fragmenting, precipitating, and resuspending the samples, at which point the samples can either be frozen for future use or hybridized to the chip. Loading chips takes nearly 2 hr, after which the samples will be hybridized overnight for 16-24 hr. On the third day, the staining and extension step takes ~4 hr. A further hour will be spent washing, coating, and drying the chips. Finally, the arrays are scanned, which may take from 15-60 min/chip, depending on the type used.
Standard laboratory safety and cleanliness precautions apply. Though the amplification is not PCR-based, separate workstations for pre- and postamplification procedures are necessary in order to minimize likelihood of contamination. The identification number of every kit-supplied reagent in use must be logged on a tracking sheet. Reagents should be thawed immediately before use and inverted several times before dispensing. The DNA to be typed must be high-quality genomic DNA (260/280 absorbance ratio of 1.6-2.0, 260/230 absorbance ratio of below 3.0), isolated by standard methods and quantified with a fluorometer. Degradation of DNA is often a contributing factor in low-quality assay results. Typically, 200 ng of DNA is required, though this amount may vary for some chip types. A Tecan liquid-handling robot can automate many steps of the protocol and minimize human error as a factor.